SPYDERSTM ON THE INTERNET
When a large earthquake occurs somewhere in the world, there is immediate interest in its details. From the simply curious lay person to the theoretical seismologist, many are fascinated by geology in violent action. Other than news reports of strong-motion effects near the epicenter, the primary way of satisfying one's interest is with seismograms recorded by well distributed seismograph stations. A decade or so ago such seismograms would not be available for months or maybe years after an earthquake occurred. No longer do we need to wait so long. Digitally recorded waveforms and a modern global communications system mean that data are now available to just about anyone on his desktop computer shortly after the shaking has stopped.
While there are a number of different sources for seismic data immediately following an earthquake, one of the most popular sources for seismograms is the Incorporated Research Institutions for Seismology (IRIS) SPYDERTM (not an acronym) system. SPYDERTM is a part of the IRIS Data Management System (DMS) designed to provide waveform data to the seismological community as quickly as possible following a large earthquake. Originally this system was called GOPHER (not an acronym) for "Go-fer the data." Because the Internet "gopher" protocol became universally known by that name in the early 1990's a new monicker was needed. The name "SPYDER" was appealing both because of the spiderlike arms representing ray paths for earthquake maps associated with an event, and because the system "spies" on interesting data all over the world. To prevent this name being usurped by yet another new Internet utility, SPYDERTM has now been registered as a trademark by IRIS.
SPYDERTM is not a real-time system that analyzes or reacts to incoming data as events occur. It is simply a system to retrieve time-windowed waveforms rapidly from a variety of digital seismographs and provide the data to anyone in the simple and popular Seismic Analysis Code (SAC) format. Begun as an experimental system in 1988 when the IRIS Data Management Center (DMC) was at interim facilities at the University of Texas, the genealogy of SPYDERTM from the original GOPHER involves too many mutations to detail here. However, the basic foundation and goals of the original system are the same today: that is, given the size and location of a recent earthquake, to recover waveforms quickly from stations of the Global Seismograph Network (GSN) and provide a simple, intuitive access technique to those data over the Internet. Then and now the SPYDERTM system receives earthquake alert messages via e-mail from the National Earthquake Information Center (NEIC) within an hour or two of any worldwide earthquake greater than magnitude 5.5. Originally the system only dialed up each telephone-equipped GSN station and downloaded selected time segments of broadband (around the predicted P wave) and long-period (from P wave to surface waves) waveform data predicted by the station-epicenter distance. The same basic data are still retrieved; however, additional data streams are acquired in special cases (very long period for earthquakes over magnitude 6.2, and low-gain channels for stations near large events), and different data retrieval techniques other than dial-up are used for different parts of the GSN. The data are presented to the interested user via an interactive menu available within a public user account on the IRIS DMC's main computer system. The menu provides the user with choices for selecting which recent earthquake is of interest, for reviewing a table of available data, for reviewing a map of stations with data, for graphically reviewing waveforms, and for downloading selected data in SAC or AH format.
After the Loma Prieta earthquake in 1989 the utility of the SPYDERTM system for providing high-quality data to the community rapidly became obvious to many, and its use increased dramatically. As more IRIS GSN stations came online with telephone access, the volume of data available increased. It soon became obvious that a single computer trying to dial up stations all over the world was not going to be able to satisfy the user needs fast enough, and the needed phone calls were going to be very expensive. About that same time the IRIS DMC moved to its present home in Seattle. In 1991 the SPYDERTM system underwent a major modification to allow for multiple computers to do the dial-up task. The system was modified to follow a client-server model where a single computer decides which time segments from which stations are of interest (the client system) and then requests server computers to actually do the dialing and downloading. The servers return the data to the client via file transfer over the Internet. There are now nine such server computers scattered all over the world, each dialing up stations that are physically close to minimize long-distance telephone charges. The primary client system at the IRIS DMC sends requests to the server systems via standard Internet e-mail based on event information received from the NEIC by e-mail.
While SPYDERTM was originally designed to acquire its data from GSN dial-up stations based on the Quanterra data logger, other parts of the GSN began to have data available in near-real time. In early 1993 the IRIS/IDA data collection center at the University of California at San Diego started providing selected waveform data from two stations with continuous telemetry from central Asia. The IRIS/IDA group wrote special software to interpret the SPYDERTM request messages and produce a data file in SPYDERTM format for inclusion in the SPYDERTM waveform archive. Since then the IRIS/IDA group has developed its own dial-up system for its stations without continuous telemetry but with dial-up telephone access. It now acts as a cooperating data center providing data based on SPYDERTM request files from nineteen stations around the world.
Also in 1993 the GEOSCOPE data center in France began providing data from some of its stations to the SPYDERTM system for events larger than magnitude 6.8. Since its system is not totally automatic, delays of a day or two between earthquake and data delivery are common.
In the spring of 1994 the U.S. National Seismograph Network (USNSN) developed an AutoDRM (see "Electronic Seismologist" in Seism. Res. Lett. Vol. 67, No. 4 for a description of the AutoDRM concept), which could be used to acquire waveform data from this satellite-telemetered network. We developed software to convert SPYDERTM requests into AutoDRM-type requests, mail these to the USNSN AutoDRM system, and then convert the returned waveform data from the GSE format (used by AutoDRM's) back to SAC for the SPYDERTM archive. Besides seismograph stations within the U.S., the USNSN also receives data from the Global Telemetered Seismograph Network (GTSN) and provides those data to SPYDERTM as if they were part of the USNSN. The Canadian National Seismograph Network uses telemetry similar to that of the USNSN and a similar AutoDRM, so data from this network are now available to SPYDERTM with virtually the same software from the Canadian network center.
Since SPYDERTM works in a client-server mode which configuration is highly flexible, there is no reason that a computer that acts as a server in some cases cannot be a client in others. For example, at Caltech a computer that acts as a server to provide waveform data from a local TerraScope station to the main SPYDERTM client machine at the IRIS DMC also acts as a client requesting data from itself and other local server machines for TerraScope data from Southern California earthquakes. In this case, and others, the requests from different clients may be intermixed on a server machine and acted on in the order received. In addition to Caltech, similar shared uses occur at the University of Washington Geophysics Program, the Earthquake Research Institute in Japan, the ORFEUS data center in the Netherlands, and the University of Alaska. A slightly different model runs at the GEOFON data center in Germany where both a SPYDERTM server and client run, but under special arrangements the client makes requests of the local server both for itself and for the main IRIS system at the DMC.
Table 1 and Figure 1 show the details of the total IRIS SPYDERTM system as of the summer of 1996, for which a total of 98 GSN stations are available. While for smaller events data are requested from only a subset of stations within a given distance, for earthquakes over magnitude 6 attempts are made to get data from all stations, shown in Figure 1. It is rare to get data from more than two thirds of these stations for any single event, since there are many possible points of failure. Besides individual stations being down, there can be problems with telephone connections, problems with remote server computers, problems at cooperating data centers over which SPYDERTM has no control, and, lately, problems with Internet connectivity due to the tremendous load its popularization is causing. SPYDERTM tries to get around some of these problems in several ways. Each SPYDERTM server will try to call its stations up to four times at varying time intervals between calls. Also, the main SPYDERTM system at the IRIS DMC will itself call stations normally accessed by remote server computers if it has not received its data over the Internet within eight hours.
Figure 2 shows a time line of the number and type of data available following a recent event. Note that there are some data available even before the official NEIC alert message goes out. This is because, under special arrangement, the IRIS SPYDERTM system receives an automatic NEIC alert generated by the USNSN real-time processing system. This early alert is used to get data from a few special GSN stations, which are sent back to the NEIC via the Internet to help them with the manual analysis of the event. The official NEIC notification e-mail is used to trigger most of the SPYDERTM stations. Note in Figure 2 that the near-real-time data from the IRIS/IDA system and the AutoDRM systems mostly arrive in groups early in the acquisition sequence, while the rest of the data (primarily from dial-up GSN stations) is scattered more smoothly over a longer time period. Some data, particularly from GEOSCOPE, may not arrive for many hours to days. Also, note that over 80% of the data which can be retrieved is available within two hours of the receipt of the official NEIC alert message.
A user accesses SPYDERTM data through the electronic bulletin board provided by IRIS at the DMC (telnet dmc.iris.washington.edu and login to account bulletin with password board). A submenu of the main menu brings up a separate interface to the SPYDERTM system which can be used directly on an ASCII terminal but which provides better and more extensive graphical displays if used through X-Windows. The data are available via anonymous FTP at the IRIS DMC for those wanting to short-cut the friendly user interface. Other SPYDERTM client systems around the world also provide a user interface which may be accessed either directly or through the IRIS DMC bulletin board. The SPYDERTM system keeps track of the amount and type of data available in the archive as well as retrieved by end users. Figure 3 illustrates the changes in some of these use statistics at the IRIS DMC over the past eight years. These statistics do not account for the large amount of data stored and accessed at other SPYDERTM sites around the world. Note that while the number of stations available and the volume of data in the archive have increased steadily over the years, the number of interactive users at the DMC peaked in 1994 and has declined some since then. This is because as the sophistication of Internet users has increased fewer need to go through an interactive menu system and more go directly to the IRIS DMC Anonymous FTP area (ftp://dmc.iris.washington.edu/pub/spyder) to retrieve the specific data desired.
With more and more data being available in essentially real time as global telemetry systems improve, will there be a need for the SPYDERTM system in the future? We plan to have SPYDERTM evolve (or mutate) additionally so it will continue to provide event-windowed waveforms rapidly in a convenient form for both seismologists and the seismically curious lay person. Besides continuing to improve the data retrieval part of SPYDERTM we hope to improve greatly the user interface, probably leaning heavily on a World Wide Web approach. We also plan to make the SPYDERTM data archive and the more permanent IRIS FARM archive fit together in a more seamless manner such that access techniques for both will be similar. SPYDERTM will be just a very rapid version of the FARM. (See "Electronic Seismologist" in Seism. Res. Lett. Vol. 67, No. 2 for more details of the FARM and other IRIS DMC data presentation techniques.)
The success of the SPYDERTM system is due in large part to the extra effort and cooperation provided by the operators of the GSN and to the cooperating data centers, particularly those listed in Table 1. The "Electronic Seismologist" (and father of SPYDERTM), as well as the many users of these data, greatly appreciate that cooperation.
SRL encourages guest columnists to contribute to the "Electronic Seismologist." Please contact Steve Malone with your ideas. His e-mail address is email@example.com.
Posted: 5 February 1999